Is the PTPase−Vanadate Complex a True Transition State Analogue

Biochemistry 2002, 41, 5865-5872

5865

Is the PTPase-Vanadate Complex a True Transition State Analogue?† Hua Deng,*,‡ Robert Callender,‡ Zhonghui Huang,§ and Zhong-Yin Zhang§ Departments of Biochemistry and Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, New York 10461 ReceiVed December 28, 2001; ReVised Manuscript ReceiVed March 8, 2002

ABSTRACT: Vanadate can often bind to phosphoryl transfer enzymes to form a trigonal-bipyramidal structure at the active site. The enzyme-vanadate dissociation constants in these enzymes are much lower than those for phosphate. Therefore, enzyme-bound vanadate moieties are often considered as transition state analogues. To test whether the enzyme-vanadate complex is a true transition state analogue beyond the simple geometry and binding affinity arguments and whether the bond orders of the VO bonds in the complex approach those of the PO bonds in the transition state, the binding properties of vanadate in the Yersinia protein-tyrosine phosphatase (PTPase) and its T410A, D356N, W354A, R409K, and D356A mutants have been studied by steady-state kinetic measurements and by difference Raman measurements. The results of the kinetic measurements show no correlation between KI and kcat or kcat/Km in these mutants. In addition, our analysis of the Raman data shows that the bond order change of the nonbridging V-O bonds in the vanadate complexes does not correlate with the kinetic parameters in a number of PTPase variants as predicted by the transition state binding paradigm. Furthermore, the ionization state of the bound vanadate moiety is not invariant across the PTPase variants studied, and the average bond order of the nonbridging V-O bonds decreased by 0.06-0.07 valence unit in the wild type and all of the mutant PTPases, either in dianionic or in monoanionic form. Thus the complex would resemble an associative transition state, contrary to the previously determined dissociative structure of the transition state. Therefore, it is concluded that vanadate is not a true transition state analogue for the PTPase reactions.

Protein-tyrosine phosphatases (PTPases)1 have an essential role in signal transduction and, together with proteintyrosine kinases, control the tyrosine phosphorylation level of proteins in the cell. The Yersinia PTPase is one member of this enzyme family. Because of its high enzymatic activity, the Yersinia PTPase has been extensively studied as a model system for all PTPases (1). The PTPases employ a common catalytic strategy in catalyzing a double-displacement mechanism in which the phosphoryl group is first transferred from the substrate to the active site Cys residue (Cys403 in the Yersinia PTPase), leading to the formation of a thiophosphoryl enzyme intermediate that is subsequently hydrolyzed by water. The invariant Arg residue (Arg409 in the Yersinia PTPase) functions in substrate binding and in transition state stabilization. The initial phosphoryl transfer step (phosphoenzyme formation) is assisted by general acid catalysis by a conserved Asp (Asp356 in the Yersinia PTPase) which protonates the leaving group. The hydrolysis of the phosphoenzyme intermediate is facilitated by the same Asp356, acting as a general base in this catalytic step (for a recent review, see ref 1). The early transition metal oxoanion vanadate is a widely used inhibitor for various phosphatases. Vanadate could † This work was supported in part by Grants GM35183 (R.C.) and CA69202 (Z.-Y.Z.) from the National Institutes of Health. * Address correspondence to this author: 718-430-2437 (phone); 718-430-8565 (fax); [email protected] (e-mail). ‡ Department of Biochemistry. § Department of Molecular Pharmacology. 1 Abbreviations: PTPase, protein-tyrosine phosphatase; pNPP, p-nitrophenyl phosphate; PTP1B, protein-tyrosine phosphatase 1B.

inhibit phosphatases by simply mimicking the tetrahedral geometry of the phosphate ion. However, in many cases, the enzyme-inhibitor dissociation constant (KI) for vanadate is much lower than that for phosphate. Because vanadate is known to be able to adopt five-coordinate structures readily, such observations have led to the hypothesis that in some cases vanadate may inhibit phosphatases by forming complexes that resemble the trigonal-bipyramidal geometry of the transition state (2, 3; cf. ref 4). In the case of PTPases, the binding affinity of vanadate for PTPases is several orders of magnitude higher than that of phosphate (5-8). To visualize directly the interactions between the PTPase active site and vanadate, the X-ray crystal structure of the Yersinia PTPase complexed with vanadate has been determined (9). The vanadate moiety in the Yersinia PTPase complex is trigonal bipyramidal, with three short equatorial nonbridging V-O bonds, one apical bridging V-O bond, and one long V-S bond to Cys403.2 Two of the equatorial V-O bonds are hydrogen bounded with two of the Arg409 side chain N-Hs as well as structural water molecules, and the remaining V-O bond is hydrogen bonded to both Val407 and Gly408 backbone N-Hs. The conserved Asp356 makes a hydrogen bond with the apical oxygen of vanadate, consistent with its role as a general acid in the phospho2 The nonbridging V-O bonds in a trigonal-bipyramidal structure refer to the bond with unprotonated oxygens, which are always on the equatorial position. In the dianionic form, all three equatorial VO bonds are nonbridging. In the monoanionic form, one of the equatorial oxygens is protonated, and the VO bond becomes a bridging V-O bond. The apical bonds are always bridging bonds since the oxygen is covalently bound to two atoms, one with V and the other with H or other atoms.

10.1021/bi016097z CCC: $22.00 © 2002 American Chemical Society Published on Web 04/06/2002

5866 Biochemistry, Vol. 41, No. 18, 2002 enzyme formation step and as a general base in the phosphoenzyme hydrolysis step. Similar results have been obtained from structures of vanadate complexes with PTP1B (10) and the low molecular weight PTPase (11). Collectively, these studies show that vanadate in the PTPase active site adopts a trigonal-bipyramidal geometry that may resemble the transition state for the hydrolysis of the phosphoenzyme intermediate in the PTPase-catalyzed reaction. These studies also support the hypothesis that vanadate inhibits PTPases by acting as a transition state analogue. The transition state binding paradigm requires that the affinity of an enzyme for a perfect transition state analogue exceed its affinity for the substrate by a factor equivalent to the increased catalytic rate in the enzymatic reaction relative to the corresponding nonenzymatic reaction (4). This latter factor has been estimated to be ∼1010- to 1011-fold for the Yersinia PTPase, based on the hydrolysis of p-nitrophenyl phosphate (12). The apparent dissociation constant (KI) for the Yersinia PTPase-vanadate complex is almost 4 orders of magnitude smaller than that of the PTPase-phosphate complex at pH 7.5 (11). However, vanadate exists mostly as tetramers and dimers in aqueous solution near neutral pH (13). Thus, the affinity of a PTPase for the monomeric vanadate adduct relative to the phosphate is likely more than 2 orders of magnitude larger than that indicated by the apparent KI values, i.e., at least 106-fold stronger than phosphate because of extensive tetra- and dimerization in solution (14, 15). Thus, the PTPase-vanadate complex is likely a plausible transition state mimic, based on both its crystal structure and a simple transition state binding paradigm argument. One goal of this study is to characterize the interaction between vanadate and PTPase quantitatively at the atomic scale. Raman difference spectroscopy (16) is well suited for the determination of high-resolution V-O bond lengths and bond orders of vanadate complexes in enzymes (17-20), and the measurements report on solution protein complexes. In such studies, the Raman spectrum of the protein-vanadate complex is measured as is that of the protein alone. The difference spectrum between these two spectra contains bands arising from the vanadate group. The vanadate bands dominate the difference spectrum due to their intrinsic large intensity and because most of the protein background spectrum subtracts out (17-19). The bond lengths and bond orders of the nonbridging V-O bonds as well as the bridging V-O bonds are determined from empirical correlations relating the vibrational stretch frequencies to these quantities (20, 21) The absolute accuracy is very high. The error in these relationships is estimated to be within (0.04 vu and (0.004 Å for bond orders and bond lengths, respectively. Moreover, the geometry of the nonbridging V-O bonds, as well as the ionic states of the bound vanadate moiety, can also be ascertained from the vibrational frequencies. The other goal is to find out if the PTPase-vanadate complex can be treated as a true transition state analogue beyond its geometry and the simple binding paradigm arguments and whether the bond orders of the VO bonds in the complex approach those of the PO bonds in the transition state. First, we argue that if PTPase-vanadate is a true transition state analogue, a simple linear correlation should exist between the binding affinities of vanadate (KI) and catalytic efficiencies (kcat and/or kcat/Km) for a series of

Deng et al. Yersinia PTPase mutants with varying degrees of activity. For a perfect transition state analogue, the correlation will have a slope of unity (22, 23). In addition, if PTPase-vanadate is a true transition state analogue, changes of the VO bonds in the complex relative to that in solution should be similar to the changes of the PO bonds in the transition state of the PTPase-catalyzed reaction. In general, a phosphoryl transfer reaction can be loosely characterized by three closely related mechanisms: dissociative, SN2-like, and associative. In a dissociative reaction, the bridging P-O bond breaking of the phosphate compound occurs well before the formation of the new P-O bridging bond in the transition state, the bond order sum of the two apical bridging bonds decreases significantly, and the total bond order sum of the equatorial nonbridging P-O bonds increases significantly compared to the ground state. In a reaction with an SN2-like transition state, the bond making and bond breaking of the apical P-O bridging bonds in the process are approximately synchronous, and the bond order sum of the corresponding nonbridging P-O bonds is not affected much relative to the ground state. In an associative transition state, the formation of a new P-O bridging bond is well advanced before the breaking of the existing P-O bond, the bond order sum of the two bridging P-O bonds is greatly increased at the transition state, and the bond order sum of the nonbridging P-O bonds is greatly decreased. Results from kinetic isotope measurements of the PTPase-catalyzed reaction as well as observations from linear free energy analysis on the catalyzed rates indicate a dissociative transition state mechanism for the Yersinia PTPase (24, 25). Furthermore, the nature of the transition state in many of the mutants under study does not change (24-27). Thus, we argue that if the PTPase-vanadate complex is a true transition state analogue, it should be similar to a certain structure of the PTPase-substrate complex along the reaction coordinate toward the transition state. This specific structure can be obtained from the kinetic parameters and the vanadate binding constant based on the quantitative application of the transition state theory. Since all PTPase variants have the same transition state as mentioned above, we can determine this specific structure for each of the PTPase variants, and it should be correlated to the bond order change of the nonbridging V-O bonds upon vanadate binding to PTPase. This method to determine if vanadate complex is a transition state analogue or not is applicable for all reaction mechanisms unless the reaction is strictly SN2-like. MATERIALS AND METHODS A 0.25 M NaH2VO4 stock solution was prepared by stirring V2O5 with 2 equiv of aqueous NaOH overnight at room temperature. p-Nitrophenyl phosphate (pNPP) was purchased from Fluka. The catalytic domains of the wildtype Yersinia PTPase (residues 163-468) and its mutants D356A, D356N, W354A, R409K, and T410A were expressed under the control of the T7 promoter in Escherichia coli BL21(DE3) and grown at room temperature after induction with 0.4 mM IPTG. All recombinant proteins were purified to greater than 95% purity as described previously (5, 12, 24, 28-30). The catalytic domain of the Yersinia

Raman Studies of the PTPase-Vanadate Complex PTPase retains the same activity as the full-length enzyme (28) and is used in all of the structural studies (9, 31-33). All enzyme assays were performed at 30 °C in either 50 mM Tris at pH 7.5 or 100 mM acetate at pH 5.5 buffers containing 200 mM KCl and 5 mM DTT. Initial rate measurements for the PTPase-catalyzed hydrolysis of pNPP were conducted as previously described (28). The Michaelis-Menten kinetic parameters were determined from a direct fit of the V vs [S] data to the Michaelis-Menten equation using the nonlinear regression program GraFit (Erithacus Software). The range of pNPP concentration used was from 0.2 to 5 Km. In all cases, the enzyme concentration was much lower than that of the substrate so that the steady-state assumption was fulfilled. Inhibition constants (KI) for the PTPase by vanadate were determined with pNPP as substrate. The mode of inhibition and the KI value were determined in the following manner. At various fixed concentrations of inhibitor (0-3 KI), the initial rate at a series of pNPP concentrations was measured ranging from 0.2 to 5 of the apparent Km values. The data were fitted to appropriate equations using kinetAsyst to obtain the inhibition constant and to assess the mode of inhibition. The enzyme samples for spectroscopic measurements were prepared by extensive dialysis in 50 mM Tris at pH 7.5 or 100 mM acetate at pH 5.5, respectively, with 200 mM KCl and 5 mM DTT, followed by concentration with a Centricon 30 centrifuge concentrator (Amicon, Lexington, MA) to the desired concentration. The concentrations of the PTPase were determined spectroscopically using a molar extinction coefficient of 16800 M-1 cm-1 at 280 nm for the catalytic domain of the Yersinia PTPase. The binary complex of the PTPasevanadate complex was prepared by adding the vanadate stock solution to the concentrated PTPase sample to achieve a molar ratio of 0.9:1. The typical concentration of the PTPase sample was about 5 mM. The Raman spectra were measured using an optical multichannel analyzer (OMA) system. The OMA system uses a Triplemate spectrometer (Spex Industries, Metuchen, NJ) with a model DIDA-1000 reticon detector connected to an ST-100 detector (Princeton Instruments, Trenton, NJ). Details of the system can be found elsewhere (16). The 514.5 nm line from an argon ion laser (model 165, Spectra Physics, Mountain View, CA) was used to irradiate the sample (∼100 mW). A half-wave retarder was used to polarize the incident light either parallel or perpendicular to the entrance slit of the spectrometer. The symmetry of a vibrational mode was assessed in terms of the ratio of Raman intensities obtained with the excitation beam in the perpendicular and parallel configurations. In this optical arrangement, the intensity ratios are 6:7 for an asymmetric mode and VO2-

W354A

R409K

D356A

945 922

915 905

920 890

915 905

934 1.605

910 1.540 -0.065 1.646 0.013

905 1.527 -0.079 1.648 0.016

910 1.540 -0.065 1.646 0.013

1.632

a νs is the symmetric nonbridging V-O stretch mode and νas1(2) is the asymmetric nonbridging V-O stretch mode. ν is the fundamental frequency, equal to the square root of average frequency squares of all νs and νas frequencies (20). SVO is the average bond order of the nonbridging V-O bonds calculated from ν using eqs 1 and 2. ∆SVO is bond order difference relative to dianionic vanadate, -VO32-, for WT, T410A and D356N and relative to monoanionic vanadate, >VO2-, for W354A, R409K, and D356A. RVO is the average bond order of the nonbridging V-O bonds calculated from ν using eqs 1 and 2. ∆RVO is the bond order difference relative to dianionic vanadate, -VO32-, for WT, T410A, and D356N and relative to monoanionic vanadate, >VO2-, for W354A, R409K, and D356A.

Accurate structural information about bonding in vanadate can be obtained from vibrational spectroscopy by using two types of empirical relationships. One is the bond length/bond strength correlation (42, 43); the other is the bond strength/ vibrational frequency correlation (20). These two correlations are given by the equations (20)

SVO ) (RVO/1.791)-5.1

(1)

SVO ) [0.2912 ln(21349/ν)]-5.1

(2)

and

where SVO and RVO are the bond order and bond length of a VO bond, respectively, and ν is the nonbridging V-O stretching frequency. The appropriate frequency to be used in eq 2 is the geometric mean of the observed nonbridging V-O stretches; ν ) [(νs2 + 2νa2)/3]1/2 (νs is the symmetric stretch and νa the asymmetric stretch) when the VO3 moiety retains nominal C3V symmetry, or, when the symmetry of the VO32- moiety is somewhat broken, like the vanadate in wild-type PTPase, [(ν12 + ν22 + ν32)/3]1/2 to determine the average bond order and bond length of the three nonbridging V-O bonds. The absolute error in these relationships in determining bond lengths and bond orders from frequency measurements is estimated to be within (0.04 vu and (0.004 Å for bond orders and bond lengths, respectively (20), and more accurate for determining bond length and bond order changes. The “fundamental” frequency of the dianionic vanadate solution model is calculated to be 869.3 cm-1 from νs ) 870 cm-1 and νa ) 869 cm-1, which yields a bond order of 1.430 vu and a bond length of 1.669 Å for each of the three nonbridging V-O bonds using eqs 1 and 2. The shift to lower frequency for the V-O bonds in the wild-type PTPase-bound vanadate translates to a somewhat lower bond order and longer bond length for the three bonds. From eqs 1 and 2, the three frequencies at 807, 848, and 876 cm-1 yield an average frequency of 848 cm-1 and an average bond order and bond length of 1.366 vu and 1.685 Å, respectively. Compared with the solution values, they change by -0.065 vu and +0.015 Å, respectively (Table 2). The fundamental frequency of the monoanionic vanadate solution model is calculated to be 933.5 cm-1 from νs ) 945 cm-1 and νa ) 922 cm-1, which yields a bond order of 1.605 vu and a bond length of 1.632 Å for each of the two

nonbridging V-O bonds using eqs 1 and 2. The shift to lower frequency for the V-O bonds in the D356A PTPasebound vanadate translates to a somewhat lower bond order and longer bond length for the three bonds. From eqs 1 and 2, the two frequencies at 905 and 915 cm-1 yield an average frequency of 910 cm-1 and an average bond order and bond length of 1.540 vu and 1.646 Å, respectively. Compared with the solution values, these parameters change by -0.065 vu and +0.013 Å, respectively, upon binding. It is interesting to note that when the active site Arg409 is mutated to Lys, the polarization of the nonbridging V-O bonds is actually increased somewhat compared with other PTPase variants (see Table 2). DISCUSSION Previous studies on RNase suggested that the RNaseuridine-vanadate complex may indeed be a true transition state analogue. The Raman studies of this complex suggested that there is a loss of 0.11-0.22 valence unit of the nonbridging P-O bonds in the transition state, based on the bond order change of the nonbridging V-O bonds in the complex (15). Studies on the secondary 18O isotope effects in the RNase-catalyzed reaction suggested that the total bond order of the nonbridging P-O bonds loses 0.130.20 valence unit in the transition state (44). However, such quantitative agreement can be interpreted as a coincidence because of limited available data. To avoid any coincidence, a number of carefully chosen PTPase variants with a wide range of catalytic efficiencies have been used in the steady-state kinetic measurements to determine if a linear correlation between the binding affinity for vanadate (KI) and the kcat or kcat/Km exists. Such correlation is expected on the basis of the transition state binding paradigm. As a whole, our data show no indication of linear correlation (Figure 1). However, when the data point from D356A is excluded from the figure, a reasonable linear correlation is observed. Since the transition state properties are the same for both D356A and D356N (25, 26, 41), there is no obvious reason to exclude this point. Thus, our results suggest that PTPase-vanadate is unlikely a true transition state analogue beyond the geometry similarity. Our Raman studies on the PTPase-vanadate complexes also support this conclusion. On the basis of the transition state binding paradigm, we argue that if vanadate moiety in this enzyme is a true transition state analogue, the nonbridg-

Raman Studies of the PTPase-Vanadate Complex ing V-O bond order change upon vanadate binding to this series of PTPase variants should reflect the P-O bond order change of the substrate in a specific structure along the reaction coordinate toward the transition state. Where this structure is located along the reaction path in the free energy diagram can be determined by using the kinetic parameters and vanadate binding constant. For a perfect transition state analogue, the logarithm of the ratio of [1/K(vanadate)]/ [1/K(substrate)] should be equal to the logarithm of the rate enhancement in the PTPase-catalyzed reaction. For a nonperfect transition state analogue, the ratio of these two values will indicate where the analogue is along the reaction pathway. This ratio was calculated for each PTPase variant and is listed in Table 1. According to this value, for wildtype PTPase, the vanadate complex would be an analogue of a state 34% on the way along the reaction path toward the transition state at pH 7.5, if it is a transition state analogue. For other PTPase variants under various conditions, the vanadate complexes would be from 23% to 45% of the way. Since previous studies have shown that the nature of the transition state in all PTPase variants is the same (2427), we expect that the nonbridging V-O bond order changes in these vanadate complexes would be correlated with these numbers. However, it is clear from Table 1 that no such correlation exists since the nonbridging V-O bond order changes are the same for all PTPase variants. In addition, if the PTPase-vanadate complex were a true transition state analogue in the sense that the nonbridging V-O bonds in the complex undergo changes similar to that of the nonbridging P-O bonds in the transition state of the PTPase-catalyzed reaction, our results would indicate an SN2like transition state, with a small associative character. Such prediction is not consistent with the results from previous studies of several types of PTPases on both phosphorylation and dephosphorylation steps, indicating a dissociative transition state (12, 24-27, 30, 38, 41). In this case, the apical bonds of the bound pentacoordinated VO32- fragment are close to zero should vanadate bind to PTPase with a structure close to that of the transition state. Hence, it is expected that the sum of the apical bonds of bound vanadate is substantially lower than the HO-VO32- bond of vanadate in solution. According to the valance bond paradigm used in the analysis here for relating bond orders (and lengths) to vibrational frequencies, the sum of the bond order about the metal atom remains constant (equal to 5 for vanadium). Hence, the finding that the total bond order of the nonbridging V-O bonds decreases by 0.195 vu (three times 0.065) for the wildtype protein (Table 2) when vanadate binds to the protein means that the sum of the bond order in the apical bonds increases by the same amount, completely contrary to the putative dissociative transition state structure. It is also contrary to the idea of a transition state mimic for its ionization state to change for the wild-type proteins and mutants assuming, reasonably, that the structure of the transition state remains the same for the various proteins (i.e., dissociative, SN2, or associative-like). Hence, the finding that the bound vanadate moiety is dianionic for wild-type PTPase and the T410A and D356N mutants, while monoanionic for the W354A, R409K, and D356A proteins, strongly argues against the idea of vanadate as a true transition state analogue. Although our results show that PTPase-vanadate is not a true transition state analogue, interesting information about

Biochemistry, Vol. 41, No. 18, 2002 5871 the interactions between vanadate and PTPase has been derived from our studies of this complex. For example, the ionic state of the bound vanadate in various mutants is determined. In the wild-type PTPase, the pKa of the -VO32group is apparently reduced by at least 3 pH units, from about pH 8.5 in solution to below pH 5.5. This is also true for T410A. In some of the mutants such as D356A, R409K, and W354A, the pKa of bound vanadate does not change from the solution value so that the bound vanadate in those mutants near neutral pH remains monoanionic. In D356N, the pKa of the bound vanadate is decreased by about 2 pH units compared to the vanadate in solution. The bond order and bond length of the nonbridging V-O bonds in the complex can be determined quantitatively from their Raman measurements. Since the V-O bond order is a direct reflection of the hydrogen-bonding strength on the V-O bond, we can also conclude that the hydrogen-bonding strength on the nonbridging V-O bonds of the vanadate moiety is not affected by D356N and T410A mutations at pH 7.5, and the hydrogen-bonding strength on the remaining two nonbridging V-O bonds does not change much when the vanadate moiety is changed from dianionic to monoanionic in D356N. Furthermore, since the KI values range from 0.24 µM in T410A to 1.23 µM in wild-type PTPase for dianionic vanadate and from 0.44 µM to 204 µM in monoanionic form, and since the interactions of the nonbridging V-O bonds do not vary much among wild-type and mutant PTPases, the large difference in the binding affinity of vanadate must be due to the differences of the V-S bond and the interaction of V-OH bridging bonds with the enzyme. Our Raman studies also show that the active site of the PTPase is capable of polarizing the nonbridging V-O bonds, no matter whether the vanadate moiety is dianionic or monoanionic. The X-ray crystallographic studies show that the polarization of the nonbridging V-O bonds is apparently facilitated by the backbone N-H groups, the side chain NHs of Arg409, and active site water molecules. It is often observed that the same type of hydrogen bonding is stronger in the hydrophobic environment than in the hydrophilic environment (45, 46). This and other observations had lead to the proposal that the coordination of the active site positively charged residues with nonbridging P-O bonds of the phosphate compounds could be stronger than that in aqueous solution in phosphoryl transfer enzyme, so that a transition state with more associative character would be favored in the enzyme compared to the solution reaction (for a brief discussion and references, see ref 47). Our results show quantitatively that the enzyme active site is capable of rendering stronger hydrogen bonding to the nonbridging P-O bonds as indicated by the 0.066 vu bond order decrease of the V-O bonds in the PTPase-vanadate complex. However, this is not happening in the transition state since it has been well established that the PTPase-catalyzed phosphoryl transfer reaction is dissociative; thus the coordination to the nonbridging P-O bonds is weakened in the transition state. Apparently, further studies are in order to solve this puzzle. REFERENCES 1. Zhang, Z.-Y. (1998) Crit. ReV. Biochem. Mol. Biol. 33, 1-52. 2. Lindquist, R. N., Lynn, J. L. J., and Lienhard, G. E. (1973) J. Am. Chem. Soc. 95, 8762-8768.

5872 Biochemistry, Vol. 41, No. 18, 2002 3. VanEtten, R. L., Waymack, P. P., and Rehkop, D. M. (1974) J. Am. Chem. Soc. 96, 6782-6785. 4. Gandour, R. D., and Schowen, R. L. (1987) Transition States of Biochemical Processes, pp 1-616, Plenum Press, New York. 5. Wu, L., Buist, A., den Hertog, J., and Zhang, Z.-Y. (1997) J. Biol. Chem. 272, 6994-7002. 6. Huyer, G., Liu, S., Kelly, J., Moffat, J., Payette, P., Kennedy, B., Tsaprailis, G., Gresser, M. J., and Ramachandran, C. (1997) J. Biol. Chem. 272, 843-851. 7. Wang, F., Li, W., Emmett, M. R., Hendrickson, C. L., Marshall, A. G., Zhang, Y.-L., Wu, L., and Zhang, Z.-Y. (1998) Biochemistry 37, 15289-15299. 8. Zhou, B., and Zhang, Z.-Y. (1999) J. Biol. Chem. 274, 3552635534. 9. Denu, J. M., Lohse, D. L., Vijayalakshmi, J., Saper, M. A., and Dixon, J. E. (1996) Proc. Natl. Acad. Sci. U.S.A. 93, 2493-2498. 10. Pannifer, A. D., Flint, A. J., Tonks, N. K., and Barford, D. (1998) J. Biol. Chem. 273, 10454-10462. 11. Zhang, M., Zhou, M., Van Etten, R. L., and Stauffacher, C. (1997) Biochemistry 36, 15-23. 12. Zhang, Z.-Y., Wang, Y., and Dixon, J. E. (1994) Proc. Natl. Acad. Sci. U.S.A. 91, 1624-1627. 13. Tracey, A. S., Gresser, M. J., and Liu, S. (1988) J. Am. Chem. Soc. 110, 5869-5874. 14. Ray, W. J., Crans, D. C., Zheng, J., Burgner, J. W., Deng, H., and Mahroof-Tahir, M. (1995) J. Am. Chem. Soc. 117, 60156026. 15. Deng, H., Burgner, J., and Callender, R. (1998) J. Am. Chem. Soc. 120, 4717-4722. 16. Callender, R., and Deng, H. (1994) Annu. ReV. Biophys. Biomol. Struct. 23, 215-245. 17. Ray, J. W. J., Burgner, I. J. W., Deng, H., and Callender, R. (1993) Biochemistry 32, 12977-12983. 18. Deng, H., Ray, W. J., Jr., Burgner, J. W., II, and Callender, R. (1993) Biochemistry 32, 12984-12992. 19. Deng, H., Wang, J., Callender, R. H., Grammer, J. C., and Yount, R. G. (1998) Biochemistry 37, 10972-10979. 20. Deng, H., Wang, J., Ray, W. J., and Callender, R. (1998) J. Phys. Chem. B 102, 3617-3623. 21. Hardcastle, F. D., and Wachs, I. E. (1991) J. Phys. Chem. 95, 5031-5041. 22. Messmore, J. M., and Raines, R. T. (2000) J. Am. Chem. Soc. 122, 9911-9916. 23. Bartlett, P. A., and Marlowe, C. K. (1983) Biochemistry 22, 4618-4624. 24. Zhang, Y.-L., Hollfelder, F., Gordon, S. J., Chen, L., Keng, Y.-F., Wu, L., Herschlag, D., and Zhang, Z.-Y. (1999) Biochemistry 38, 12111-12123.

Deng et al. 25. Hengge, A. C., Sowa, G. A., Wu, L., and Zhang, Z.-Y. (1995) Biochemistry 34, 13982-13987. 26. Hoff, R. H., Wu, L., Zhou, B., Zhang, Z.-Y., and Hengge, A. C. (1999) J. Am. Chem. Soc. 121, 9514-9521. 27. Hoff, R. H., Hengge, A. C., Wu, L., Keng, Y.-F., and Zhang, Z.-Y. (2000) Biochemistry 39, 46-54. 28. Zhang, Z.-Y., Clemends, J. C., Schubert, H. L., Stuckey, J. A., Fisher, M. W. F., Hume, D. M., Saper, M. A., and Dixon, J. E. (1992) J. Biol. Chem. 267, 23759-23766. 29. Zhang, Z.-Y., Wang, Y., Wu, L., Fauman, E., Stuckey, E., Schubert, H. L., Saper, M. A., and Dixon, J. E. (1994) Biochemistry 33, 15266-15270. 30. Zhang, Z.-Y., Wu, L., and Chen, L. (1995) Biochemistry 34, 16088-16096. 31. Schubert, H. L., Fauman, E. B., Stuckey, J. A., Dixon, J. E., and Saper, M. A. (1995) Protein Sci. 4, 1904-1913. 32. Fauman, E. B., Yuvaniyama, C., Schubert, H. L., Stuckey, J. A., and Saper, M. A. (1996) J. Biol. Chem. 271, 18780-18788. 33. Stuckey, J. A., Schubert, H. L., Fauman, E., Zhang, Z. Y., Dixon, J. E., and Saper, M. A. (1994) Nature 370, 571-575. 34. Wilson, E. B. J., Decius, J. C., and Cross, P. C. (1955) Molecular Vibrations, McGraw-Hill, New York. 35. Deng, H., Kurz, L., Rudolph, F., and Callender, R. (1998) Biochemistry 37, 4968-4976. 36. Keng, Y.-F., Wu, L., and Zhang, Z.-Y. (1999) Eur. J. Biochem 259, 809-814. 37. Zhang, Z.-Y., Palfey, B. A., Wu, L., and Zhao, Y. (1995) Biochemistry 34, 16389-16396. 38. Zhao, Y., and Zhang, Z.-Y. (1996) Biochemistry 35, 1179711804. 39. Denu, J. M., and Dixon, J. E. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 5910-5914. 40. Zhao, Y., Wu, L., Noh, S. J., Guan, K.-L., and Zhang, Z.-Y. (1998) J. Biol. Chem. 273, 5484-5492. 41. Hengge, A. C., Zhao, Y., Wu, L., and Zhang, Z.-Y. (1997) Biochemistry 36, 7928-7936. 42. Brown, I. D. (1992) Acta Crystallogr. B48, 553-572. 43. Brown, I. D., and Wu, K. K. (1976) Acta Crystallogr. B32, 1957-1959. 44. Sowa, G. A., Hengge, A. C., and Cleland, W. W. (1997) J. Am. Chem. Soc. 119, 2319-2320. 45. Jeffrey, G. A., and Saenger, W. (1991) Hydrogen Bonding in Biological Structures, Springer-Verlag, Berlin. 46. Vinogradov, S. N., and Linnel, R. H. (1971) Hydrogen Bonding, Van Nostrand Reinhold Co., New York. 47. O’Brien, P. J., and Herschlag, D. (1999) J. Am. Chem. Soc. 121, 11022-11023. BI016097Z